On the potential of jointly-optimised solutions to spoofing attack detection and automatic speaker verification